Using human computation in dead-zone based 2D pattern matching

This paper examines the application of human computation (HC) to two-dimensional image pattern matching. The two main goals of our algorithm are to use turks as the processing units to perform an efficient pattern match attempt on a subsection of an image, and to divide the work using a version of dead-zone based pattern matching. In this approach, human computation presents an alternative to machine learning by outsourcing computationally difficult work to humans, while the dead-zone search offers an efficient search paradigm open to parallelization—making the combination a powerful approach for searching for patterns in two-dimensional images.